0 votes
in Azure by

What is single point of failure -- is a part of the system, if it fails, will stop the entire system from working.

1 Answer

0 votes
by

The infrastructure of each Azure data center is divided into multiple sections which are treated as fault domains. These sections of the infrastructure (which are not necessarily individual servers or server racks) are designed in such a way that a failure of one fault domain is extremely unlikely to affect any other fault domain. When you deploy a service, the Azure Fabric Controller automatically locates the roles in at least two different fault domains so that a failure in one domain will not affect all instances of your service.

This is the reason why Azure recommends you to have 2 instances so that Azure fabric will place your application role instances in more than one fault domain to avoid failure.

Upgrade Domain -

It is a logical unit of grouping the role instances and it does not exist physically. When we have 2 instances of role running then automatically each of the deployment or instances are treated as different upgrade domain. So when an upgrade of guest OS, host OS or application update need to be performed then only one upgrade domain based instances is updated while upgrade domain 2 based instance keep serving the user's requests. Once upgrade of first instance is completed then second upgrade domain based instance gets updated and so on. So at any point of time during upgrade at least one instance is alive and serving the users request hence you don't face the downtime of your application.

Related questions

0 votes
asked Mar 10 in Azure by Robindeniel
0 votes
asked Nov 12, 2022 in Operating System by john ganales
...